Dear colleagues, in partnership with Theologus at home, we welcome Dr. Gregory Marcus in San Francisco, California, USA, and Dr. Lian in Munich, Germany, for an update on atrial fibrillation with two very important studies published by our two guests very recently: DECAF on coffee consumption in patients with atrial fibrillation, and the Munich study on new-onset atrial fibrillation after coronary artery bypass grafting. Dear Dr. Hammer, we are very happy to host for the first time a heart surgeon at home, and dear Greg, welcome back at Marcus at home after our 2022 webinar on alcohol and coffee intake in atrial fibrillation. Happy to have you back.
Thanks for joining also from my side. As a brief introduction, I think Dr. Marcus doesn't need much of that. He's an endowed professor in atrial fibrillation research, probably the leading working group when it comes to the effects of caffeine and alcohol on atrial fibrillation, and recently published studies on that. That's where we come together here. Also, a study that was presented at the American Heart Association meeting. Professor Marcus is Associate Chief of Cardiology for Research at the University of California, San Francisco Health Institute, and he's Associate Editor at JAMA as well. As I said, his research interest is atrial fibrillation, and he's conducted numerous trials on that. We are very happy to have you here again and hear some of your thoughts on the most recent one, the DECAF study, and also maybe hear your comments which you published as an editorial on the study by Florian Hammond and Dr. Dr. Herman. Thank you for coming. We are very happy to have a heart surgeon in this round for the first time. Dr. Hammerman has quite some research achievement and won prizes and got research funding from the KUNAR Forinius, and recently last year was awarded a PRIME Advanced Clinician Scientist Booster Award to facilitate his research next to his clinical duties. We are very keen to hear your presentation on the CABG-AF study on atrial fibrillation post-bypass surgery. So without further ado, I would ask you, Dr. Marcus, first to start with your presentation. The floor is yours.
Thank you. Again, a pleasure to be here and talk about our study, and I look forward to talking with Dr. Herman about his study. This one is our Does Eliminating Coffee Avoid Fibrillation or DECAF trial that we were privileged to present at the AHA and recently published. Of course, conventional wisdom is that coffee caffeine leads to more tachyarrhythmias. This is what we were at least frequently taught in medical school, and many patients are warned against consuming caffeine, especially those with fast heart rhythm. Our journey here began several years ago where we conducted this observational study using data from the UK Biobank. We just looked at how much caffeinated coffee hundreds of thousands of people in the UK consumed at a baseline state and followed them over time to see what their risk was of various tachyarrhythmias. This is a figure showing all of those tachyarrhythmias combined, and we were quite surprised to find this very consistent inverse relationship. So the more caffeinated coffee an individual described consuming at baseline, the lower the risk of incident tachyarrhythmias. When we teased out what fast rhythms were really contributing here, it was primarily atrial fibrillation. We recognize that as an observational study, this is of course prone to confounding. So maybe there's something different about the type of people who decide to consume caffeinated coffee that's the true causal operator. We then thought about how we can apply randomized control trial study designs that we know are the best way to mitigate against confounding. This led to our CRAVE trial: Coffee and Real-time Assessment of Atrial and Ventricular Ectopy. I enjoyed describing or discussing with you all on this forum a few years ago where this was just a day on versus off randomized caffeinated coffee consumption. Our primary outcome was the frequency of premature atrial contractions, which we know are a really potent predictor of atrial fibrillation, and we found no relationship between caffeinated coffee consumption and PACs. Now importantly, as a kind of broad-strokes big picture context limitation or consideration related to our DECAF trial, I will get back to in a moment. We did find more premature ventricular contractions on days when people were randomly assigned to caffeinated coffee. So that is certainly worthy of consideration, especially for people who suffer from PVCs, and certainly also worthy of further investigation as that was a secondary endpoint. So then of course we wanted to know, okay, how about atrial fibrillation itself? We conducted this DECAF trial. Chris Wong, who was a fellow with us in cardiac electrophysiology, now actually already full professor at the University of Adelaide, began work with me on this when he was a fellow with us. He subsequently went to Adelaide. We were able to enroll 200 patients with persistent atrial fibrillation from UCSF, University of Adelaide, and Chris Chung, also a former UCSF fellow now in Canada. We only enrolled people undergoing electrical cardioversion or scheduled for electrical cardioversion of atrial fibrillation, and we did this for a couple reasons. One, it assured that their atrial fibrillation was clinically relevant. It was important enough that some treating physician decided you need a procedure for your AFib. We need to get you out of atrial fibrillation. Two, it gave us a common starting point where everyone was starting in sinus rhythm. They didn't need to actually undergo the electrical cardioversion because we know sometimes people are started on antiarrhythmic drugs and they show up in sinus rhythm, and we didn't want to exclude them. Another important inclusion criterion was people had to have consumed caffeinated coffee at some regular time in the last 5 years. So they didn't need to be consuming caffeinated coffee when they were enrolled because we know a lot of patients stop consuming caffeinated coffee, not because it caused a problem, but because their treating physician instructed them to do so, and we wanted to make sure to include them. They couldn't have had recent CT surgery, which is obviously relevant to Dr. Herman's study he will tell us about, nor could they have had a recent ablation or plan to change their management in the next 6 months. They were randomly assigned to try to consume a cup of caffeinated coffee a day at least, you could consume more if you want, versus try to avoid all coffee and all caffeine for six months. Our primary outcome was a recurrence of clinically relevant atrial fibrillation, meaning they needed to have a diagnosis of atrial fibrillation recognized by a treating provider and documented with some ECG evidence of atrial fibrillation including from a wearable monitor. These are the results. On the left is what happened to actual coffee consumption. Both groups started, and you can see these kind of dark lines in these box plots are the medians. Interquartile range are the boxes, and then the larger range in the y error bars. But if you focus in on the medians, it's about a cup a day. In the group randomly assigned to coffee consumption, median amounts remained about a cup a day. There were some outliers in those randomized to abstinence. Sometimes people did consume some coffee, but the median was actually zero cups of coffee in follow-up. This was ascertained at 1 month, 3 months, and 6 months and remained fairly consistent. In our main analysis, primary outcome, we found that those randomly assigned to caffeinated coffee abstinence actually experienced a significantly higher risk of AFib recurrence. So those randomly assigned to caffeinated coffee consumption appear to experience a lower risk of incident atrial fibrillation over those six months. Those are all my slides, and I look forward to discussing more with all of you.
Well, maybe give the word to Dr. Herman first if you like. We have some particular questions maybe.
Thanks. I read the publication online and I was impressed. It was a very interesting trial, and that's definitely shown by the amount of views it's already received, almost 150,000 views, which is spectacular. It's very interesting. It's also related to cardiac surgery a little bit because it's the same recommendation that we frequently make as well about caffeine, and I think a study in this area might be interesting as well in the future because there's no actual data that's related to caffeine intake after cardiac surgery. I haven't found any yet. We still make the recommendation that patients shouldn't be drinking caffeinated coffee after surgery. So that's one interesting thought. I saw that looking deeper into your data, I saw that the proportion of patients in each group that had a certain amount of specific medications were very similar. Is that something that you stratified in randomization?
We did not. We stratified by site. Use of medications may have traveled a little bit by site. We were fortunate that randomization thankfully kind of balanced that out. It's almost perfectly balanced in the groups.
And my second question is about the topic clinically relevant AF because I did a lot of research in this study on subclinical AF, which I didn't expect before, but it developed. The more data came in, you saw it was much more subclinical AF than clinical AF. How would that be something that you're interested in, also with caffeinated drinks?
Yes. It's a limitation of the study, but also I think just important in understanding the implications. It was designed to be quite pragmatic. We would have loved to have put loop recorders in everyone as in your study, but we didn't have the resources in this particular study. I will say this study was also quite difficult to convince patients to participate in. If you look at the consort diagram, there's a pretty big drop off between the thousands of people we talked to compared to those that ultimately were randomized. The most common reason being many people didn't want to give up their coffee. In fact, we initially when we started, we had hoped to do this for a year and then realized people are really just not going to do it for a year. So let's truncate it to 6 months, which thankfully was sufficiently powered. Some number just didn't ever drink caffeinated coffee. But yes, the clinical relevance did mean they had to present with atrial fibrillation, have a diagnosis of atrial fibrillation, and so almost certainly there's subclinical atrial fibrillation going on. It would be very interesting to determine that. To put on a really skeptical lens, it would be not unreasonable to say, well, maybe the people who were consuming caffeinated coffee just felt a little better and they were less likely to complain to their doctors.
That's the exact thought I had.
Yeah. So that could be the case. And then it becomes an issue of what is the research question and what is a mediator versus a confounder. So if your outcome is healthcare utilization for AFib, it still is reducing that. It's just the mechanism by which it's doing it is not necessarily suppressing the AFib itself. We actually think, you know, a common question I get is, does this even make sense mechanistically from a biological standpoint? Does it make sense that caffeine or coffee might suppress AFib? There are actually several potential biologically plausible mechanisms. One is that most atrial fibrillation is probably vaguely mediated, and interestingly, presumably the post-cardiac surgery may not fall into that category, but the more spontaneous forms are. So maybe the sympathetic effects of caffeine actually help suppress AFib. Caffeine blocks adenosine. We know from our experience in the EP lab, adenosine triggers AFib. So maybe it's protecting it that way. Animal models have shown that caffeine tends to prolong left atrial refractoriness. That is an electrophysiologic property that has been shown to prevent AFib. That's how sotalol works. Dofetilide, amiodarone are probably suppressing AFib by prolonging left atrial refractoriness. Then, kind of a little bit longer term, there's very good epidemiologic evidence, although not randomized control trial evidence, that people that consume caffeinated coffee experience a lower risk of diabetes, and for reasons we don't completely understand, diabetes predicts AFib even after adjusting for sedentary behavior and BMI. Finally, there's good data such as from Prash Sanders' group that physical activity can reduce AFib. And we showed in our CRAVE trial that when people were wearing Fitbits, when they're randomly assigned to caffeinated coffee, they exercise more. So maybe that's a pathway that might lead to a lower risk of AFib. So it's not as if we are free of potential mechanisms, and the finding does fit with observational evidence. But to be scientific, we have to always put on our skepticism hats, and that's how we arrive at the truth. We chip away at things and consider all the possibilities.
Before we give the word to Dr. Herman for his presentation, one question I have: obviously we have to rely on the statements of the patients concerning their coffee use. I was so impressed in your alcohol study that you provided people with a wrist alcohol measurement tool, which is obviously not available for coffee. Is there any humoral factor that we could measure, any toxicological screening that was not only in your study? I understand.
Yeah, I think if we had serum samples, theoretically we could have. There is not, you're right. We have an ongoing study using alcohol. The probably most reliable now is an ankle-worn sensor, which then brings up issues of stigma and things like that. But yes, we've been very interested to try to prove the exposure. Our suspicion is that if self-report was inaccurate, presumably it would be on the side of drinking more than they acknowledged, which then presumably would have more reduced our power rather than created a spurious false positive. In CRAVE, we did a few things. We used our geolocation tracking where we showed that people visited coffee shops more often when they were randomized to coffee. We had them press a button on the Ziopatch when they consume coffee to try to mitigate against recall bias. And in CRAVE, we told all our participants, 'We'll pay for all of your caffeinated coffee as long as you give us a receipt with a date on it, whether you were supposed to or not.' And we did find that actually people seemed to be quite compliant. There is this thing called the Hawthorne effect, this effect where if someone knows they're being watched, they're more likely to adhere. Maybe we didn't have that in DECAF, but that is another important potential limitation. Certainly, these are all great ideas for future research.
